Video-driven three-dimensional motion hip joint impingement syndrome auxiliary diagnosis method

Abstract

The invention relates to a hip joint diagnosis index automatic quantitative calculation method based on three-dimensional equation expression. The method comprises the following steps: step 1, processing a CT image through a boundary-based contour line extraction method to obtain a three-dimensional hip joint point cloud; 2, establishing a three-dimensional equation expression of the hip joint, and calculating diagnosis indexes, namely an acetabulum index and a femoral head protrusion index; step 3, extracting a human body action sequence from the motion video by using a rapid monocular three-dimensional body action capture method; and 4, calculating five key points and the like through the three-dimensional coordinates of the pelvic bone center, the femoral head center, the femoral neck center and the lesser trochanter vertex. According to the method, an integrated auxiliary diagnosis motion model is constructed, the auxiliary diagnosis motion model avoids errors caused by distortion of muscles and skin in motion based on rigid characteristics of hip joint bones, two FAI auxiliary three-dimensional diagnosis indexes are provided based on morphological characteristics, errors caused by view angle changes in a CT image are avoided, and the accuracy of diagnosis is improved. And a reliable hip joint impact diagnosis basis is provided.

Description

Technical Field

[0001] The present invention belongs to the interdisciplinary field of computer graphics and clinical medicine, and particularly relates to a three-dimensional hip joint motion model for assisting in the diagnosis of femoroacetabular impingement syndrome by combining CT images and patient motion videos. Background Art

[0002] Femoroacetabular impingement syndrome (FAI) mainly refers to the abnormal contact or collision between the proximal femur and the acetabular rim due to anatomical abnormalities of the femoral head or acetabulum, which in turn causes damage to the cartilage and acetabular labrum and leads to hip pain. Therefore, timely diagnosis is crucial for providing appropriate treatment or preventive measures to patients. With the increasing prevalence of hip-related problems, improving diagnostic assistance methods has become particularly important for ensuring timely and effective patient care, preventing further joint damage, reducing pain, and improving the quality of life of patients.

[0003] Currently, there are mainly two clinical diagnostic methods: one is based on medical images to calculate relevant indicators, and the other is based on motion analysis and dynamic test results.

[0004] The method based on medical images mainly involves the quantitative calculation of the relative anatomical relationship between the proximal femur and the acetabular rim. This quantitative analysis makes the diagnosis more objective and accurate. Existing studies include three-dimensional quantitative assessment and alpha angle measurement from the femoral head-neck junction based on magnetic resonance imaging, as well as geometric models for automatically calculating three-dimensional quantitative FAI indicators on CT (Computed Tomography) images. Although these methods provide objective diagnostic indicators, they only reflect the skeletal morphology of the femoral head and acetabulum, do not consider the actual skeletal collision situation in dynamic motion, and lack intuitiveness and comprehensiveness.

[0005] The other method based on motion analysis and dynamic testing considers the dynamic information of ligaments and skeletal structures during movement. Doctors diagnose by directly observing and evaluating the range of motion of the hip joint and pelvic tilt, and these evaluations are usually based on actions such as hip flexion, abduction, and external rotation. These clinical diagnostic methods rely entirely on doctors' observations and patients' pain feedback, and have a high degree of subjectivity. Furthermore, some portable hardware devices are designed to estimate the spatio-temporal parameter changes of the patient's gait before and after surgery to quantify gait changes. However, this method requires the patient to wear corresponding devices, the analysis results depend on the accuracy of the devices, and doctors cannot observe the anatomical structure.

[0006] In the existing technologies for auxiliary diagnosis by combining motion capture, special motion capture devices are required to obtain the motion data of patients, and then quantitative indicators are calculated. In actual clinical diagnosis, the operability is relatively low, and there is a lack of intuitive motion visualization and analysis. In another existing technology, optical motion capture is combined with nuclear magnetic resonance images to track and visualize hip joint motion. However, this existing technology requires markers to be attached to the patient's skin surface, which is prone to inaccuracy due to skin deformation, so it lacks practicality and effectiveness.

[0007] In view of this, by combining the anatomical information in CT images with the pose information in videos, a video-driven comprehensive three-dimensional hip joint motion model for auxiliary diagnosis is proposed. Different from the method that relies on skin surface markers, rigid bones are directly used for motion capture and reconstruction. In addition, three-dimensional FAI diagnostic indicators are objectively defined through morphological analysis, providing a quantitative evaluation of the motion model. Summary of the Invention

[0008] The present invention proposes an automatic quantitative calculation method for hip joint diagnostic indicators expressed by three-dimensional equations, and a method for generating a video-driven three-dimensional hip joint motion model for FAI auxiliary diagnosis.

[0009] The automatic quantitative calculation method for hip joint diagnostic indicators expressed by three-dimensional equations includes the following steps:

[0010] Step 1: Process the CT image through a boundary-based contour extraction method to obtain a three-dimensional hip joint point cloud;

[0011] Step 2: Establish a three-dimensional equation expression of the hip joint and calculate the diagnostic indicators Acetabular Index (AI) and Femoral Head Coverage (FHC);

[0012] Step 3: Use a fast monocular three-dimensional body motion capture method to extract the human motion sequence from the motion video;

[0013] Step 4: Calculate five key points through the three-dimensional coordinates of the pelvic centroid, femoral head center, femoral neck center, and apex of the lesser trochanter, including the hip joint center point, left upper femoral endpoint, right upper femoral endpoint, left lower femoral endpoint, and right lower femoral endpoint;

[0014] Step 5: Generate a three-dimensional hip joint motion model through motion redirection technology;

[0015] Step 6: Combine the diagnostic indicators with the visualization of the motion model for auxiliary diagnosis.

[0016] Further, in step 1, for the hip joint CT image sequence of the patient, segmentation and contour extraction are performed by the region growing method, and a three-dimensional point cloud of the hip joint is obtained by interval sampling, including the point clouds of the pelvis, left femur, and right femur.

[0017] Further, in step 2, a three-dimensional equation expression of the hip joint is established and diagnostic indices, namely the acetabular index and the femoral head protrusion index, are calculated. The specific steps are as follows:

[0018] Step 2.1 First, based on the fitting algorithms of a basic spherical surface, Gaussian surface, and elliptical curve, the hip joint point cloud data is fitted to establish equation expressions for the femoral head, femoral neck, the intersection line of the head and neck, and the outer edge line of the acetabulum. According to the position of the femoral head on the femur, a segmentation plane is determined using the "bounding box method", the point set belonging to the femoral head is intercepted, and the spherical parameter equation is obtained through the spherical fitting algorithm. The equation parameters are solved using the least squares method.

[0019] Step 2.2 The femoral neck contour line is extracted according to the single-layer sphere center intercept method, and the center of the femoral neck is determined. A point set on the femur within a certain distance range is selected, and the femoral neck equation expression is obtained through the elliptical curve fitting algorithm and the least squares method. The solution methods for the equation expressions of the intersection line of the head and neck and the outer edge line of the acetabulum are the same as those for the femoral neck.

[0020] Step 2.3 The morphological definitions of the two diagnostic indices, AI and FHC, are determined, and their specific positions and corresponding calculation methods in the hip joint are clarified. AI is calculated through vector relationships. Based on the line connecting the vertex of the Y-shaped cartilage and the most prominent point of the acetabulum, and the line connecting the vertices of the Y-shaped cartilage on both sides, the inclination angle of the acetabulum is determined to quantitatively analyze the risk of hip joint impingement. FHC measures the femoral head coverage rate by calculating the femoral head protrusion index. Using the spherical equation of the femoral head and the straight line equation connecting the center of the outer edge line of the acetabulum and the center of the femoral head, the geometric relationship between the femoral head and the acetabulum is determined to obtain the protrusion degree. Through the femoral neck equation expression, the specific values of these two indices can be calculated, providing a basis for the quantitative analysis of the hip joint.

[0021] Further, in step 3, a fast monocular three-dimensional body motion capture method is used to extract the human body motion sequence: a single camera is used to capture the human body motion video, and then the motion skeleton is extracted from the video and converted into a three-dimensional human body posture, so as to realize the motion capture of the patient's hip. Taking the monocular video as the input, the posture parameters of the body are output through feature extraction and three-dimensional posture regression, so as to realize the estimation and extraction of the human body motion posture, and obtain the human body motion sequence extracted from the video. Among them, the five key points related to the hip motion are the center point of the pelvis (Pelvis), the left hip point, the right hip point (Left Hip, Right Hip), and the left knee point, the right knee point (Left Knee, Right Knee).

[0022] Further, in step 4, the three-dimensional coordinates of the centroid of the pelvis, the center of the femoral head, the center of the femoral neck, and the apex of the lesser trochanter are used to calculate the five key points of the center point of the hip joint, the upper end point of the left femur, the upper end point of the right femur, the lower end point of the left femur, and the lower end point of the right femur, which specifically include:

[0023] Calculate the 5 key points corresponding to the hip joint part of the human body model extracted from the video, namely the Pelvis point, the LeftHip point, the Right Hip point, the Left Knee point, and the Right Knee point. The center point of the hip joint corresponds to the Pelvis point in the human body model, and the centroid of all points of the pelvis is taken as the coordinates of this point;

[0024] The upper end points of the femur correspond to the Left Hip point and the Right Hip point. Based on the geometric relationship between the femoral head and the femoral neck, the center line of the head and neck is determined, and the coordinates of the upper end point are obtained through the intersection point with the spherical surface of the femoral head;

[0025] The lower end points of the femur correspond to the Left Knee point and the Right Knee point. The vertex of the lesser trochanter is selected as the lower end point. By intercepting the point set and using a quadratic polynomial model to fit the surface of the lesser trochanter, the point with a derivative value of 0 is finally calculated as the key point at the lower end of the femur. Through the equation expressions of the femoral neck, the intersection line of the head and neck, and the outer edge line of the acetabulum in step 2.2, the precise coordinates of these five key points are further calculated.

[0026] Further, in step 5, the three-dimensional hip joint motion model is generated by the motion retargeting technology: based on the key points of the hip joint model calculated in step 4, the three-dimensional hip joint model obtained in step 1 is skeleton-bound, and the motion retargeting technology is adopted to reconstruct the hip joint motion using the motion information in the human body motion sequence extracted in step 3.

[0027] Further, the auxiliary diagnosis by combining the diagnostic indicators with the visualization of the motion model in step 6 also includes visually displaying the diagnostic indicators obtained in step 2 and the three-dimensional hip joint motion model obtained in step 5 simultaneously, and realizing the auxiliary diagnosis of FAI through the combination of quantitative calculation and qualitative analysis.

[0028] Compared with the prior art in this technical field, the superior effects of the present invention are as follows:

[0029] 1. The auxiliary diagnosis method for hip impingement syndrome driven by video for three-dimensional motion of the present invention constructs an integrated auxiliary diagnosis motion model. The auxiliary diagnosis motion model is based on the rigid characteristics of the hip joint bones, avoiding errors caused by the distortion of muscles and skin during motion, and combining the quantitative analysis of diagnostic indicators, making the diagnosis of FAI more accurate and detailed.

[0030] 2. The auxiliary diagnosis method for hip impingement syndrome driven by video for three-dimensional motion of the present invention proposes two FAI auxiliary three-dimensional diagnostic indicators based on morphological features. Through the calculation method derived from the three-dimensional model, it avoids errors caused by changes in viewing angles in CT images and provides a reliable basis for diagnosing hip impingement.

[0031] 3. The auxiliary diagnosis method for hip impingement syndrome driven by video for three-dimensional motion of the present invention is an auxiliary diagnosis tool integrating automatic quantitative calculation and motion visualization. The auxiliary diagnosis tool for motion visualization can more intuitively evaluate FAI, improving the diagnosis efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Schematic diagrams of five key points related to hip movements involved in the method of the present invention;

[0033] Figure 2 Flowchart of the auxiliary diagnosis method for hip impingement syndrome based on video-driven three-dimensional hip joint motion of the method of the present invention;

[0034] Figures 3a - 3b Schematic diagram of the morphological FAI evaluation index of the method of the present invention;

[0035] Figure 4 Schematic diagram of the femoral head fitting of the method of the present invention;

[0036] Figures 5a - 5b Curve graph of femoral characteristics obtained based on parametric equations of the method of the present invention;

[0037] Figure 6 Schematic diagram of the vertex of the Y-shaped cartilage shown by the method of the present invention;

[0038] Figure 7Illustrates the three-dimensional acetabular angle calculation method for the method of the present invention;

[0039] Figure 8 Illustrates the calculation method of the protrusion index for the method of the present invention;

[0040] Figure 9 Schematic diagram of the positioning of the upper femoral endpoints based on the spherical equation for the method of the present invention;

[0041] Figures 10a - 10b Schematic diagram of the automatic positioning of the lesser trochanter based on the quadratic equation for the method of the present invention;

[0042] Figures 11a - 11b -11c Schematic diagram of the matching method between the motion skeleton and the human skeleton for the method of the present invention;

[0043] Figures 12a - 12b Schematic diagram of the automatic detection result of the impact area for the method of the present invention;

[0044] Figure 13 Schematic diagram of the real-time visualization display combining the diagnostic index and the motion model for the method of the present invention. Detailed implementation manners

[0045] In order to be able to more clearly understand the above objects, features and advantages of the present invention, the following combines the attached Figures 1 - 13 and specific implementation manners to further describe the present invention in detail.

[0046] Before introducing the embodiments of the method of the present invention, briefly describe the morphological definitions of the FAI diagnostic indices AI and FHC, and further describe them using unified mathematical symbols to facilitate the understanding of the embodiments of the method of the present invention.

[0047] Morphological definitions of FAI evaluation indices:

[0048] The acetabular angle refers to the angle formed by the straight line passing through the vertices of the Y-shaped cartilage of the bilateral acetabula (also known as the "H" line) and the line connecting the lowest edge of the ilium and the outermost edge of the acetabulum in the anteroposterior X-ray film of the bilateral hip joints, as Figure 3a shown.

[0049] The femoral head protrusion index, also known as the femoral head coverage rate and acetabular coverage rate, refers to the percentage by which the femoral head protrudes beyond the acetabular coverage, and can be calculated by the ratio of the area covered by the acetabulum to the total surface area of the femoral head or by the ratio of line segment lengths, as Figure 3b shown.

[0050] Morphological definitions of key points:

[0051] The hip joint center corresponds to the Pelvis point in the human body model extracted from the video. Although the pelvis has slightly different shapes and positions in different individuals, its centroid is usually located between the two iliac bones, which can effectively represent the overall position of the hip joint. Therefore, the centroid of all points of the pelvis is taken as the Pelvis point.

[0052] The upper end point of the femur corresponds to the Left Hip / Right Hip in the human body model extracted from the video. The upper end of the femur is located at the connecting part of the femur and the pelvis and on the spherical surface of the femoral head. Therefore, the line connecting the center of the femoral head and the center of the femoral neck on the same side is determined to obtain the center line of the head and neck, and the point farther from the center of the femoral neck among the intersection points of this line and the spherical surface of the femoral head is taken as the Left Hip / Right Hip.

[0053] The lower end point of the femur corresponds to the Left Knee / Right Knee in the human body model extracted from the video. Considering that this key point in the video is at the knee position, while the CT image only extends to the middle of the femoral shaft, the smaller conical elevation below the connection of the femoral head and the femoral neck, that is, the vertex of the lesser trochanter, is selected as the Left Knee / Right Knee.

[0054] Hypotheses for establishing an equation expression for the three-dimensional structure of the hip joint:

[0055] I. Since the femoral head is almost spherical anatomically, a sphere equation is used to represent the femoral head.

[0056] II. The shape of the femoral neck, the intersection line of the femoral head and neck, and the outer edge line of the acetabulum are all approximately elliptical contour lines. Therefore, the center of the ellipse is used as the density center.

[0057] Embodiment

[0058] The method described in the present invention includes the following steps:

[0059] S1: Processing the CT image by a boundary-based contour extraction method to obtain a three-dimensional hip joint point cloud

[0060] For the CT image sequence of the patient's hip joint, it is segmented and the contour line is extracted by the region growing method, and a three-dimensional point cloud of the hip joint is obtained by its interval sampling, including the pelvis left femur and right femur and other parts, where N, M, S represent the number of points in the point cloud sets of the corresponding parts;

[0061] S2: Establishing a three-dimensional equation expression of the hip joint and calculating the diagnostic indicators AI and FHC

[0062] The three parts of the pelvic bone, left femur, and right femur in the hip joint three-dimensional point cloud are further subdivided according to the geometric shape. The femur is divided into the femoral head (fitted with a sphere), the femoral neck, and the intersection line of the head and neck (fitted with an elliptical curve). The outer edge line of the acetabulum in the pelvic bone is fitted with an elliptical curve to complete the establishment of the equation expression, and then the auxiliary diagnostic indicators AI and FHC are calculated:

[0063] S21: Fitting of the femoral head

[0064] Use the left femoral head point set P l For example (the fitting of the right femoral head is the same). For the CT images of different patients, a fixed ratio cannot be directly determined to construct the cutting plane. Therefore, the "bounding box method" is used to define a plane to separate the femoral head from other parts of the femur. Based on the anatomical structure of the human hip joint, a cuboid where the femoral head is located is extracted by setting a threshold. According to the position of the diagonal interface of the cuboid, such as Figure 4 the cutting plane in, select the rectangular bounding box, and the plane where it is located is used to separate the femoral head. The specific method is as follows:

[0065] Calculate the maximum and minimum values of the femoral head in three dimensions (i.e., the x, y, and z directions) and combine them to obtain two sets of coordinates as (x max , y max , z max ) and (x min , y min , z min ). Using the calculated maximum and minimum coordinate values, take the coordinate range close to the top of the femoral head. The x value range is [x min +(x max -x min )*0.6, x max , the y value range is [y min , y max , and the z value range is [z min +(z max -z min )*0.75, z min +(z max -z min )*0.92]. Combine the boundary values of the ranges of x, y, and z to form 8 points (each of x, y, and z has 2 boundary values, a total of 2*2*2 = 8 combinations) to obtain a cuboid, as shown in Figure 4 . Use the plane defined by the bounding box method to cut the femoral head, and take the points located above the plane as the point set H for fitting the femoral head l :

[0066] According to Hypothesis I, establish the femoral head equation:

[0067] (x - x head )2 +(y - y head ) 2 +(z - z head ) 2 =R head 2 #(1)

[0068] Set the loss function according to the least - squares method:

[0069]

[0070] Substitute the point set H l ={(x1,y1,z1),(x2,y2,z3),...,(x i ,y i ,z i ),...} into formula (2) and minimize the objective function formula (3) to solve for the coordinates C head =(x head ,y head ,z head ) of the femoral head center and the femoral head radius R head which are respectively the following formulas (3) - (4):

[0071]

[0072] where d0 is a constant term with the value of formula (5):

[0073]

[0074] S22: Femoral neck fitting

[0075] According to Hypothesis II, the femoral neck contour line is an elliptical curve. First, use the single - layer sphere - center intercept method to extract the femoral neck contour line and determine the femoral neck center. First, take the points on the femur whose distance from the femoral head center does not exceed R head + ε to construct the point set M neck near the femoral neck contour, where set the offset ε ∈ [3mm, 5mm]. Calculate the coefficients H0 and the constant term b0 of the normal equation according to the average value of the point set M neck . Solve the normal equation to obtain the coefficients a0, a1, a2. Map the point set to the plane l1: a0x + a1x + a2 = 0 to obtain the point set M neck-proj . Use the least - squares method to fit an ellipse to obtain the ellipse equation and calculate the center C neck =(x neck ,y neck ,z neck ) coordinates, as the following formulas (6) - (7):

[0076] Ax 2 +Bxy + Cy 2 +Dx + Ey + F = 0 #(6),

[0077]

[0078] S23: Fitting of the head - neck intersection line

[0079] Utilize the characteristic that the Gaussian curvature changes significantly near the head - neck intersection line of the femoral head. First, calculate the Gaussian curvature of each point on the femur, and take the points with significant Gaussian curvature change as the point set for fitting the head - neck intersection line

[0080] Then, use a method similar to that for fitting the femoral neck. For example, as in formulas (6) and (7), calculate the coefficients H' and the constant term b0' of the normal equation system according to the average value of the point set G, solve the normal equation system to obtain the coefficients a0', a1', a2', map the point set to the plane l2: a0'x + a1'y + a2' = 0, and use the least - squares method to fit the head - neck intersection line equation, and calculate the center C of the head - neck intersection line hn =(x hn , y hn , z hn ), as in formulas (8) - (10):

[0081] A'x 2 +B'xy + C'y 2 +D'x + E'y + F' = 0 #(8),

[0082]

[0083] Thus, the parametric equation formula (10) of the head - neck intersection line is obtained:

[0084]

[0085] Among them, c and d represent the major axis and minor axis of the head - neck intersection line of the femoral head respectively, and the values are as in formula (11):

[0086]

[0087] The parameter t ∈ (-π, π) represents the angle at different positions on the curve. For example, t = 0 corresponds to the position of the point on the curve closest to the inner side of the body (only for the left femur, and the right femur is symmetric to the left femur), θ is the rotation angle of the ellipse relative to the x - axis. When A' = C', t takes Otherwise, take The fitting result is as Figure 5a shown;

[0088] S24: Fitting of the acetabular rim line

[0089] Connect the center C of the femoral head head and the points on the femoral head to form a ray, calculate the distance from each point on the pelvis to the ray, and the points with a distance less than the threshold (the threshold is taken as 0.2) are used to construct the acetabular point set Through the centroid, the center of the femoral neck, and the angles formed by each point on the acetabulum and the center point of the femoral neck, through formula (12):

[0090]

[0091] where is and C neck at M ac the vector formed at each point, is the center P of the M ac point set c and C neck the vector formed, and the point set with the largest angle is selected as the fitting point set M m ={m i m}, for the obtained point set, use the ellipse model to establish the equation of the outer edge line of the acetabulum, and use the least squares method to solve for the center C ac =(x ac ,y ac ,z ac ) of the outer edge line of the acetabulum, and the parametric equation Eq2 of the outer edge line of the acetabulum, as shown in formula (13):

[0092]

[0093] where the parameter is t'∈(-π,π), c' and d' are the major and minor axes of the curve respectively, a0”, a1”, a2” are the three parameters of the projection plane, and θ' is the rotation angle of the ellipse relative to the x-axis (corresponding to a0', a1', a2' and θ in the head-neck intersection line), and the fitting result is as Figure 5b shown;

[0094] S25: AI calculation process

[0095] Connect the center point C of the outer edge line of the acetabulum ac and the center C of the femoral head head to obtain the parametric equation Eq3 of the line: [x,y,z]=C head +t*(C ac -C head ), calculate the distance from each point in the acetabular point set M ac to the ray Among all the points with a distance less than the threshold (the threshold is taken as 1), select the point closest to C head to obtain the vertex Y_point of the Y-shaped cartilage, that is As shown in Figure 6 ;

[0096] Substitute the corresponding t value into Eq2 (Formula 13) (t = π for the left femur and t = 0 for the right femur), obtain the most prominent point at the top of the acetabulum, and finally calculate the acetabular angle size through vector relationships. Taking the left side as an example, as shown in Formula (14):

[0097]

[0098] Where l A is the connection line of the Y-shaped cartilages on both sides, and l B is the connection line between the vertex of the Y-shaped cartilage on the same side and the most prominent point of the acetabulum. The position is as shown in Figure 7 ;

[0099] S26: FHC calculation process

[0100] According to the femoral head-neck intersection line equation Eq1 and the acetabular outer edge line equation Eq2, take the corresponding t value to obtain the positions b and c on the outer sides of the two curves. Combine the straight line where points b and C head are located and the femoral head spherical surface equation, and calculate the coordinates of the other intersection point a of the straight line passing through point b and the femoral head center with the femoral head spherical surface. As shown in Figure 8 ; finally, calculate the femoral head protrusion index through vector relationships, as shown in Formula (15):

[0101]

[0102] Where represents the unit vector perpendicular to the upward direction (i.e., the unit vector in the Z-axis direction [1, 0, 0]), and represent the vectors of the two line segments respectively;

[0103] S3: Extraction of human motion sequences based on a fast monocular 3D body motion capture method

[0104] Adopt a fast monocular 3D body motion capture method to achieve motion capture of the patient's hip. Using the single-person motion video of the patient as the input, T is the video length. Extract features from each frame of the video, and obtain the SkinnedMulti-Person Linear (SMPL) model parameters of each frame through parameter regression, obtaining the motion sequence of 3D human pose estimation. The SMPL model consists of the shape parameter β ∈ R 10 and the pose parameter θ ∈ R 24×3 which are composed. Among them, the pose parameter θ ∈ R 24×3 represents the axis-angle expression of the three-dimensional rotation angles of the 24 joint points of the human body relative to their parent nodes. Given a video sequence, calculate and obtain For the subsequent step S6, it is transformed into a rotation matrix representation \(N\in\mathbb{R}\). 3×3 Since the present invention focuses on hip movement postures, only the rotation matrices of five key points (Pelvis, Left Hip, Right Hip, Left Knee, and Right Knee) near the hip relative to their parent nodes are used. As shown in Figure 1 , at the \(t\)-th frame, the rotation matrix of node \(i\) relative to its parent node in the current coordinate system is denoted as \((N i ) (t) \in\mathbb{R} 3×3 , as shown in Equation (16):

[0105]

[0106] where \(I\in\mathbb{R} 3×3 is the identity matrix, and is the axis-angle representation of the three-dimensional rotation angle of node \(i\) relative to its parent node at the \(t\)-th frame. Thus, the rotation matrices of the five joint points of the patient's hip relative to their parent nodes are obtained;

[0107] S4: Calculation of key points based on the three-dimensional coordinates of the pelvic centroid, femoral head, femoral neck center, and lesser trochanter vertex

[0108] Corresponding to the 5 key points of the hip joint part of the human body model extracted from the video, the hip joint center, the upper endpoints of the left and right femurs, and the lower endpoints of the left and right femurs are selected on the three-dimensional point cloud reconstructed from the CT image for subsequent motion reconstruction work;

[0109] S41: Hip joint center

[0110] It corresponds to the Pelvis point in the human body model extracted from the video. Although the shape and position of the pelvis vary slightly among different individuals, its centroid is usually located between the two iliac bones and can effectively represent the overall position of the hip joint. Therefore, the centroid of all points of the pelvis is taken as the Pelvis point;

[0111] S42: Upper endpoint of the femur

[0112] It corresponds to the Left Hip / Right Hip in the human body model extracted from the video. The line connecting the center of the femoral head and the center of the femoral neck on the same side is determined to obtain the head-neck center line, and the point farther from the center of the femoral neck among the intersection points of this line and the spherical surface of the femoral head is taken as the Left Hip / Right Hip;

[0113] Connect the femoral head center \(C head and the femoral neck center \(C neck obtained in S21 and S22 to get the head-neck center line, solve the intersection points of the head-neck center line and the spherical surface of the femoral head, and obtain the coordinates of the upper endpoint of the femur. The results are as shown in Figure 9 .

[0114] S43: Lower end point of femur

[0115] Corresponding to Left Knee / Right Knee in the human body model extracted from the video. Considering that this key point in the video is at the knee position, while the CT image only extends to the middle of the femoral shaft, a smaller conical elevation below the connection of the femoral head and femoral neck, that is, the vertex of the lesser trochanter, is selected as Left Knee / Right Knee;

[0116] Similar to the case of intercepting the points on the spherical surface of the femoral head when fitting with the S21 femoral head, the maximum and minimum values of the femoral point set F in three dimensions are calculated. According to the position of the lesser trochanter on the femur, a plane is used to intercept the points on the surface of the lesser trochanter ( Figure 10a ) to obtain the point set Use the quadratic polynomial model to establish the surface equation Eq3 of the lesser trochanter, that is, formula (17),

[0117] A”x 2 +B”y 2 +C”xy + D”x + E”y + F” = 0 #(17),

[0118] Fit the surface of the lesser trochanter by the least squares method, as Figure 10b shown. Calculate the derivative values of each point on the fitted surface, and take the points with derivative values of 0 as the extreme points on the fitted surface, that is, the key points at the lower end of the femur;

[0119] S5: Generation of a three-dimensional hip joint motion model based on motion retargeting technology

[0120] S51: Calculation of bone binding and rotation matrix

[0121] First, based on the five key points calculated in S4, perform bone binding on the three-dimensional hip joint model and correspond it to the bones in the human motion model extracted from the video. The bone naming and corresponding relationships are as Figure 11a and 11b shown:

[0122] The motion capture file usually describes the direction in the local coordinate system. The local coordinate system of the bone starting from the i-th node is denoted as l, and its origin is at the i-th node, and the Y-axis points to the end node, as Figure 11c shown, and the global coordinate system is denoted as g;

[0123] To convert the rotation matrix of the node to the global coordinate system, use the following formula (18):

[0124]

[0125] where the parent bone node is n - 1, and n = 0 represents the pelvic bone node, Denote the rotation matrix of node k in the local coordinate system;

[0126] To calculate the rotation difference matrix between the source skeleton and the target skeleton of the i-th node Use formula (19):

[0127]

[0128] where N i ∈R 3×3 and N' i ∈R 3×3 are the rotation matrices of the source skeleton and the target skeleton of the i-th node in the static posture in the global coordinate system respectively.

[0129] S52: Generation of the three-dimensional motion model of the hip joint

[0130] Use the coordinate sequence of node i in each frame to describe the motion of the hip joint. Since the motion of the hip joint is rigid, translation and scaling are ignored during the calculation, and homogeneous coordinates are used for overall transformation. Calculate the coordinates of node i at the t-th frame through formula (20):

[0131]

[0132] where O i (0) ∈R 1×3 is the global coordinate of node i in the static posture, is the rotation matrix of the source skeleton of node i at the t-th frame, and thus the three-dimensional motion model of the hip joint is obtained.

[0133] S6: Auxiliary diagnosis by combining the diagnostic index and the visualization of the motion model

[0134] For the three-dimensional motion model of the hip joint obtained in S5, the collision area detection results between the acetabulum and the left and right femurs are determined by the Boolean operation between the meshes, and these areas are visualized, as Figure 12a and 12b shown. The system integrates the relevant technologies of S1 - S6, and can realize the real-time visualization display schematic diagram of combining the diagnostic index and the motion model, as Figure 13 shown.

[0135] The present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only to illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims.

Claims

1. A method for automatic quantitative calculation of hip joint diagnostic indicators based on three-dimensional equation expression, characterized in that: The following steps are involved: Step 1: Process the CT image by using the boundary-based contour extraction method to obtain a three-dimensional hip joint point cloud; Step 2: Establish a three-dimensional equation expression of the hip joint and calculate the diagnostic indicators acetabular index and femoral head protrusion index; Step 3: Extract human motion sequences from sports videos using a fast monocular 3D body motion capture method; Step 4: Calculate five key points through the three-dimensional coordinates of the pelvic centroid, femoral head center, femoral neck center, and lesser trochanter vertex, including the hip joint center point, left femoral upper end point, right femoral upper end point, left femoral lower end point, and right femoral lower end point; Step 5: Generate a three-dimensional motion model of the hip joint through motion redirection technology; Step 6: Combine diagnostic indicators with visualization of motion models for auxiliary diagnosis.

2. The method for automatically calculating the diagnostic indexes of hip joints based on the three-dimensional equation expression according to claim 1 is characterized in that: In step 1, the patient's hip joint CT image sequence is segmented and contour lines are extracted by region growing method, and the three-dimensional point cloud of the hip joint is obtained by interval sampling, including the point clouds of the pelvis, left femur and right femur.

3. The method for automatically calculating the diagnostic indexes of hip joints based on the three-dimensional equation expression according to claim 1 is characterized in that: In step 2, a three-dimensional equation expression of the hip joint is established and the diagnostic indicators acetabular index and femoral head protrusion index are calculated, which specifically includes the following steps: Step 2.1 First, based on the basic spherical, Gaussian and elliptic curve fitting algorithms, the hip joint point cloud data is fitted to establish equation expressions for the femoral head, femoral neck, head-neck intersection line and acetabular outer edge line. According to the position of the femoral head on the femur, a segmentation plane is determined using the bounding box method, and the point set belonging to the femoral head is intercepted and the spherical parameter equation is obtained through the spherical fitting algorithm. The equation parameters are solved using the least squares method; Step 2.2: Extract the contour of the femoral neck and determine the center of the femoral neck according to the single-layer sphere center intercept method, select a set of points on the femur that are within a certain range of distance from the femoral neck and solve it through the elliptic curve fitting algorithm and the least squares method to obtain the equation of the femoral neck. The equation expression and solution method of the head-neck intersection line and the acetabulum outer edge line are also the same as those of the femoral neck. Step 2.3 determined the morphological definitions of the two diagnostic indicators, AI and FHC, and clarified their specific positions in the hip joint and the corresponding calculation methods. AI was calculated through vector relationship, based on the line connecting the apex of the Y-shaped cartilage and the most protruding point of the acetabulum, as well as the line connecting the apex of the Y-shaped cartilage on both sides, to determine the inclination angle of the acetabulum and quantitatively analyze the risk of hip joint impingement. FHC measured the femoral head coverage by calculating the protrusion index of the femoral head. The geometric relationship between the femoral head and the acetabulum was determined using the spherical equation of the femoral head and the straight line equation connecting the center of the acetabulum outer edge line and the center of the femoral head, and the degree of protrusion was obtained. The specific values ​​of the two indicators could be calculated through the femoral neck equation expression, providing a basis for quantitative analysis of the hip joint.

4. The method for automatically calculating the diagnostic indexes of hip joints based on the three-dimensional equation expression according to claim 1 is characterized in that: In step 3, a fast monocular 3D body motion capture method is used to extract human motion sequences: a single camera is used to capture human motion video, and then the motion skeleton is extracted from the video and converted into a 3D human posture to achieve motion capture of the patient's hip. The monocular video is used as input, and the body posture parameters are output through feature extraction and 3D posture regression to achieve estimation and extraction of human motion posture, and obtain a human motion sequence extracted from the video, wherein the five key points related to hip motion are the pelvic center point, left hip point, right hip point, left knee point, and right knee point.

5. According to claim 1, the five key points of the hip joint center point, the upper end point of the left femur, the upper end point of the right femur, the lower end point of the left femur, and the lower end point of the right femur are calculated by the three-dimensional coordinates of the pelvic centroid, the femoral head center, the femoral neck center, and the lesser trochanter vertex in step 4, specifically including: Calculate the five key points corresponding to the hip joint of the human model extracted from the video, namely the Pelvis point, the Left Hip point, the Right Hip point, the Left Knee point and the Right Knee point. The center point of the hip joint corresponds to the Pelvis point in the human model, and the center of mass of all the points of the pelvis is taken as the coordinate of the point. The upper end point of the femur corresponds to the Left Hip point and the Right Hip point. Based on the geometric relationship between the femoral head and the femoral neck, the center line of the head and neck is determined, and the coordinates of the upper end point are obtained through the intersection with the spherical surface of the femoral head; The lower end point of the femur corresponds to the Left Knee point and the Right Knee point. The vertex of the lesser trochanter is selected as the lower end point. By intercepting the point set and fitting the lesser trochanter surface using a quadratic polynomial model, the point with a derivative value of 0 is finally calculated as the key point of the lower end of the femur. The precise coordinates of the five key points are further calculated through the equation expressions of the femoral neck, the head-neck intersection line, and the acetabulum outer edge line in step 2.

2.

6. The method for automatically calculating the diagnostic indexes of hip joints based on the three-dimensional equation expression according to claim 1 is characterized in that: Step 5 describes generating a three-dimensional motion model of the hip joint by motion redirection technology: based on the key points of the hip joint model calculated in step 4, the three-dimensional hip joint model obtained in step 1 is skeletally bound, and motion redirection technology is used to reconstruct the hip joint motion using the motion information in the human motion sequence extracted in step 3.

7. The method for automatically calculating the diagnostic indexes of hip joints based on the three-dimensional equation expression according to claim 1 is characterized in that: The auxiliary diagnosis by combining diagnostic indicators with visualization of motion models described in step 6 also includes visualizing the diagnostic indicators obtained in step 2 and the three-dimensional motion model of the hip joint obtained in step 5 at the same time, and realizing auxiliary diagnosis of FAI by combining quantitative calculation and qualitative analysis.

Images